Hits:
Indexed by:期刊论文
Date of Publication:2017-10-01
Journal:APPLIED THERMAL ENGINEERING
Included Journals:Scopus、SCIE、EI
Volume:125
Page Number:480-488
ISSN No.:1359-4311
Key Words:Inverse heat conduction problem; Levenberg-Marquardt algorithm; Conjugate gradient method; Least-squares method; Temperature-dependent thermal conductivity; Thermal protection system
Abstract:Metallic materials such as an Inconel and an alloy steel play very important roles for bearing in the reusable metallic thermal protection system (TPS) for a hypersonic aircraft. Accurate determination of temperature-dependent thermal conductivities of these metallic materials is a key issue for both design and optimization of the TPS, which determines transient temperature field and further thermal stress distribution. However, it is very difficult to directly measure these temperature-dependent thermal conductivities with high temperatures, or analytically calculate them. Inverse problems provide new insights for accurately determining temperature-dependent thermal conductivities of these materials with high service temperatures, by using additional 'temperature measurements. In the present work, multi, parameters of temperature-dependent thermal conductivities of an Inconel in a reusable metallic TPS are simultaneously estimated by solving, a transient nonlinear inverse heat conduction problem (IHCP). The Levenberg-Marquardt (LM) algorithm, the conjugate gradient method (CGM) and the Least-squares (IS) method are employed for the solution. The accuracy, the efficiency, the robustness and the convergence stability of these three methods for estimating temperature-dependent thermal conductivities of the Inconel are investigated in detail. A relaxation factor is introduced into a LM algorithm for the first time, attributed to which the convergence stability of the LM algorithm is improved. Finally, the optimal choice of the three methods for simultaneously estimating multi-parameters of temperature dependent thermal conductivities of the Inconel is suggested, that is accurate, robust, convergence stable and the most efficient. (C) 2017 Published by Elsevier Ltd.